CT extension: distance to parallel AC conductors

[ben]

Can you use your fluke to check AC millivolts at the end of the CT then at the end of your extension? That'll show whether or not you're getting some signal loss thru the extension, they should be around the same.

[peppersass]

Some interesting measurements. There may be some attenuation, but I think the main problem is noise and SNR.

Also, my Fluke 189 might have some problems. Earlier I found that it under-reads the house voltage by about 1 volt compared with my Fluke 101, my oscilloscope and the GEM. The 189 is pretty old and may need calibration.

Here are the measurements:

Image 8-18-21 at 8.22 PM.jpeg (113.86 KiB) Viewed 3265 times

I measured voltage across each of the two CTs at the subpanel (disconnected from the extensions) and at the GEM (reconnected to the extensions and disconnected from the GEM.) I made measurements with the HPHWH in standby and running. I measured with a Fluke 189 and Fluke 101. The designations CT 2 and CT 4 refer to the HPHWH breaker numbers in the subpanel.

As you can see, the standby voltage at both ends fluctuates -- a lot. It's quite rapid and the numbers provided are just the lowest and highest numbers I saw when staring at the meter for 15 seconds or so. Would a rapidly fluctuating low-level signal like this cause the GEM to report 0W when the reading should be about 4W -- which is what I calculated from the reading of the ammeter in line with one of the HPHWH leads at the subpanel breaker -- ?

The running voltage fluctuates, too, but mostly in the 1/100ths place, sometimes in the 1/10ths place. It's not enough fluctuation to significantly affect the running wattage reported by the GEM, but it may be enough to account for the discrepancy between what the GEM shows (about 430W) and the ammeter calculation (about 450W). That is, unless that discrepancy is just the power factor difference.

The Fluke 101 readings seem more credible. The standby voltage is about 1/100th of the running voltage, which correlates with my calculations based on ammeter readings at the HPHWH breaker of 4W-5W in standby and 450W running. Also, the Fluke 189 reads higher running voltages at the GEM end, which doesn't make sense. I think the 189 needs calibration and/or repair.

There could be some attenuation by the extensions. The max standby voltage readings from the Fluke 101 are lower at the GEM. There doesn't appear to be any attenuation of the higher voltage when running.

Given that the signals from the CTs at the subpanel are fluctuating when the CTs are disconnected from the extensions, the noise must not be coming from AC conductors that are parallel to the extensions. I'm wondering if it could be coming from a poorly-designed switching power supply in the HPHWH that runs the low-voltage DC electronics. I'll have to put my oscilloscope on the AC lines to check that.

Note that I haven't tried a noise jumper on the lower two ports of the HPHWH GEM channel yet. I figured when the standby wattage went to zero when I grounded the CAT cable shield drains the noise (which probably comes from parallel AC conductors) had been taken care of. I'll try a jumper when I get a chance.

[ben]

You could move the CTs over to a known constant load (something purely resistive: lighting, an outlet you can plug a hair dryer/etc. into, anything similar) and measure that as you should have a pretty constant voltage in that case. A 1000W hair dryer should be 1000W at the GEM. If it isn't you have a problem there.

[peppersass]

The CTs are in a subpanel with only two breakers at present: the well pump and the HPHWH. No outlets available, though I guess I could wire one up.

I have another idea. I'll put a pair of Split 60s on the breaker in the main panel that feeds the subpanel. That will eliminate the extensions. The load will include the well pump, but it only goes on a few times a day at most, and it's separately monitored by the GEM through the extensions so I can subtract any well pump power from the readings at the main panel subpanel breaker.

This should let me see several things at both ends of the extensions: 1) Standby watts being drawn by the HPHWH, 2) Running power drawn by the HPHWH, and 3) Whether the .2Wh spikes are really happening.

Of course, if it's noise from the HPHWH switching power supply gumming up the works, that'll probably show up at the main panel subpanel breaker, too. I'll have to use my oscilloscope to confirm that.

[peppersass]

Early report: With the HPHWH running, the Split 60s at the main panel / GEM read 6W higher than the Micro 40s in the subpanel at the other end of the extensions. Only 1.8% difference, which I suppose could be the difference in CT accuracy, but I suspect it's resistive loss. If that 6W loss is constant, not proportional, then it's enough to zero out the HPHWH's standby power.

[peppersass]

After monitoring for about a day, I'm still not seeing the regular 0.2kWh spikes the HPHWH seems to think it's consuming. I don't see them on the micro 40s monitoring the breaker and I don't see them on the Split 60s in the main panel monitoring the feed to the subpanel.

The difference in consumption between the two channels is about 6%, with the CTs on the extensions being lower, which seems too high to be caused by CT accuracy differences. I tried moving the CT and twisted pair lines from the terminal strips to clamp-on splices, but that didn't make an appreciable difference.

I don't think capacitance in the twisted pairs would cause attenuation, do you? The cable is more than 10 years old, but it was unused and should be fine. It's actually flooded buriable cable, quite rugged. It's probably CAT5, being that old.

I'm going to have to take a look with the oscillscope. I might also try running a temporary extension of CAT6 along the floor, which would be at least 7 feet from any parallel AC conductors.

All that said, I very much doubt that noise or attenuation is causing the GEM to miss the 0.2kWh spikes. That's got to be a bug in the HPHWH, no?

[peppersass]

Another update: I took a closer look at the Tesla Gateway charts and I don't see the 0.2kWh spikes there, either. I also examined the 5-minute data downloaded from the Gateway and I don't see them there, either. Overnight we tend to idle at about 320W, and all the increases above that correlate with the HVAC fan and refrigerators cycling. It's a little hard to tell because the HPHWH chart resolution is very poor -- hard to tell exactly when the spikes supposedly occurred -- but I'm convinced they're not there.

[ben]

Another update: I took a closer look at the Tesla Gateway charts and I don't see the 0.2kWh spikes there, either. I also examined the 5-minute data downloaded from the Gateway and I don't see them there, either. Overnight we tend to idle at about 320W, and all the increases above that correlate with the HVAC fan and refrigerators cycling. It's a little hard to tell because the HPHWH chart resolution is very poor -- hard to tell exactly when the spikes supposedly occurred -- but I'm convinced they're not there.

Was reading this blog here: https://tinkertry.com/replaced-my-gas-w ... 020-update

They were having issues in 2020 with the app not calculating properly.